摘要
研究一种关于隐马尔可夫模型的多序列比对,利用值和特征序列的保守性,通过增加频率因子,改进传统隐马尔可夫模型算法的不足。实验表明,新算法不但提高了模型的稳定性,而且应用于蛋白质家族识别,平均识别率比传统隐马尔可夫算法提高了3.3个百分点。
A new multiple sequence alignment about Hidden Markov Models(HMMs) is researched,using the conservative feature of L value and consensus sequence,by increasing frequency factor,traditional HMMs learning algorithm is improved.Experiment indicates that not only the stability of the model is improved,but also a average improvement of 3.3% is achieved for protein family recognition by comparing the new algorithm with the traditional one.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第7期171-174,共4页
Computer Engineering and Applications
基金
国家"十一五"科技支撑计划重大项目资助No.2006BAJ05A06
重庆市科委自然科学基金(No.2007BB2205)
重庆市科委重点攻关项目(No.2008AC0043)~~
关键词
隐马尔可夫模型
多序列分析
蛋白质识别
hidden markov models
multiple sequence analysis
protein recognition